AI’s contribution to the climate transition

Artificial intelligence (AI) has the potential to accelerate the climate transition by supporting mitigation, adaptation and resilience across five key impact areas: Transforming complex systems; innovating technology discovery and resource efficiency; nudging behavioural change; modelling climate systems and policy interventions; and managing adaptation and resilience.

AI applications can improve energy system integration, enhance efficiency, and optimise investment in sustainable infrastructure, particularly in emerging markets. AI is already being used to forecast energy supply and demand, identify suitable areas for renewable energy investment, and support smart infrastructure development.

Transforming complex systems

Decarbonising the global economy requires systemic changes to sectors such as energy, transport, cities, and land. AI enables improved integration and operation of these systems by processing real-time data and supporting decision-making. Examples include AI-driven grid management for renewables, and smart urban planning tools such as Singapore’s Smart Nations initiative. AI also improves access to climate finance in emerging markets by reducing information asymmetries and increasing risk prediction accuracy. This helps attract investment into sustainable projects where capital is most needed.

Innovating technology discovery and resource efficiency

AI accelerates the development and commercialisation of low-carbon technologies. DeepMind’s GNoME identified over two million new crystal structures, potentially enhancing energy storage and renewable generation. AI also supports breakthroughs in protein discovery (e.g., AlphaFold) that could improve alternative protein (AP) development. In industry, AI increases resource efficiency and productivity. For example, Amazon’s optimisation of packaging logistics has saved over three million tonnes of packaging material. AI also improves recycling rates through computer vision systems like those used by GreyParrot.

Nudging and behavioural change

AI can influence consumer choices to reduce greenhouse gas emissions, which could account for a 40–70% reduction by 2050. Tools such as Google Nest and Oracle’s Opower use behavioural science and AI to promote energy-efficient decisions. In the food sector, Winnow Vision helps reduce food waste through automated tracking, while Google Maps promotes fuel-efficient travel routes in mobility. These applications demonstrate AI’s capacity to tailor recommendations and address behavioural barriers to climate-friendly choices.

Modelling climate systems and policy interventions

AI improves climate modelling and forecasting. For instance, IceNet, developed by the British Antarctic Survey and the Alan Turing Institute, enhances sea ice predictions. AI is also used to evaluate the effectiveness of climate policies. Climate Policy Radar applies AI to analyse thousands of policies and identify best practices. Additionally, AI can support the integration of “Beyond GDP” metrics into economic modelling, allowing policymakers to consider sustainability, wellbeing, and equity alongside traditional economic growth.

Managing adaptation and resilience

AI contributes to early warning systems and disaster preparedness. Google’s FloodHub can forecast floods five days in advance in over 80 countries. AI-based digital twins such as NVIDIA’s Earth-2 are being developed to improve extreme weather prediction. It also enables long-term resilience planning by simulating ecosystem changes and tracking environmental risks, including biodiversity loss and water stress.

Quantifying AI’s impact on emissions reduction

The report estimates that AI could reduce annual emissions by 3.2–5.4 GtCO₂e by 2035 across power, food, and transport sectors. In energy, AI improves grid efficiency and renewables integration (1.8 GtCO₂e). In food, it enhances adoption of APs, reducing emissions by up to 3.0 GtCO₂e. In mobility, AI supports shared transport and electric vehicle uptake, contributing 0.5–0.6 GtCO₂e in reductions. These reductions exceed the estimated 0.4–1.6 GtCO₂e increase from AI-related data centre emissions, supporting a net benefit from AI deployment.

Conclusion

AI offers significant potential to support the net-zero transition. However, markets alone may not guide AI towards the most impactful and equitable applications. Active government involvement is necessary to ensure AI is used sustainably and inclusively. Public investment, regulatory oversight, and support for AI adoption in developing countries are recommended to maximise climate and economic benefits.